This contribution builds on current trends in dairy cattle breeding and discusses continuing and new avenues in genomics, phenomics, breeding and molecular genetics. Last decade has seen an immense growth in genome-wide genotyping with SNP arrays, which has enabled breeders and producers to quickly genetically improve their herds. The number of genotyped animals is now in the millions and it is possible that in the near future all breeding, if not all production, dairy animals will be genotyped. At the same time genome re-sequencing is becoming cheaper. Appropriate sequencing strategies and imputation will give us whole-genome sequence for the millions of genotyped animals. Investment in sequencing is not there yet, but some breeding programs are already sequencing all AI bulls. Although SNP array information is sufficient for genomic selection within breeds, sequence information will facilitate studies and breeding across breeds through increased mapping resolution. Sequence data will provide information on short insertions and deletions and large-scale structural variants. New data formats will be required, such as genome graphs. Modeling the vast amounts of data will pose a significant challenge. Recent studies on population and evolutionary genetics in human proposed a data structure (based on evolutionary trees with recombination) that enables almost perfect data compression. I will show how to use this data structure for quantitative genetic modeling of haplotype effects, either with SNP array or sequence data. When applied to a genome region the model has a sparse structure similar to the pedigree model. Such a model could facilitate within- and across-breed data analysis. The volume of phenotype data will also increase through the use of existing and new recording technologies, for example, activity sensors, automatic milking systems, infrared spectroscopy, gene-expression, etc. These high-throughput phenotypes are often high-dimensional and can be used to infer various properties of an animal with deep learning methods. For example, routine infrared spectroscopy data from milk samples is used in the UK to predict cow pregnancy status, which is in turn used as phenotype data for genetic evaluation of fertility. The sheer volume of these data will pose a question of how different breeding entities and activities contribute to genetic gain. Blockchain-like technologies could be used to track the flow of information in genetic evaluations and flow of genome segments in a breeding program to create a transparent tracking system. Such a system is required as current public-private breeding landscape drives rapid changes in genetic diversity. Introduction of genomic selection increased the loss in genetic diversity despite the potential to enlarge the breeding pool. I will show with simulation of a phenotypic and genomic dairy breeding program that optimal contribution selection should be used to increase the efficiency of converting genetic diversity into genetic gain. Finally, molecular genetics will drive new developments via genome editing, gene drives and potentially also cell-line breeding.